Stereo vinyl records
-
Vinyl records, as well as 78s (with the exception of a very few, and incompatible, records from the very first years), have a V-formed track where the stylus rides. On a stereo record, the sound from the left channel is engraved in the 'hillside' that is the one arm of the V, the right channel sound is engraved in the opposite V arm hillside. The question: How is the sound 'encoded'? If the sound in both channels is identical, like mono sound (or the source located exactly midway between the microphones), do the hillsides then move out to each side, and come in from both sides, in identical moves? If so, the bottom of the V, the valley between the hillsides, forms a straight line, but moving up and down, and so does the stylus tip. Or, do the two hillsides move in parallel to the same side and back again, 'holding hands'? Then the stylus tip moves from side to side, but not up and down? On a 'true' monophonic record, the V track goes from side to side, with constant depth, suggesting that for a stereo record, the second alternative is the right one, but I am far from 100% sure about that. Can anyone tell, with absolute certainty, which alternative is right for a stereo record? Does the stylus tip go in a straight line, but up and down, or from side to side, but at the same vertical depth, when a stereo record plays exactly the same sound in both speakers?
trønderen wrote:
The question: How is the sound 'encoded'?
Ben Krasnow (hardware engineer over at [Google research](https://en.wikipedia.org/wiki/Verily)) covered this on his Youtube channel. [Electron microscope slow-motion video of vinyl LP - YouTube](https://www.youtube.com/watch?v=GuCdsyCWmt8&t=306s)
-
Vinyl records, as well as 78s (with the exception of a very few, and incompatible, records from the very first years), have a V-formed track where the stylus rides. On a stereo record, the sound from the left channel is engraved in the 'hillside' that is the one arm of the V, the right channel sound is engraved in the opposite V arm hillside. The question: How is the sound 'encoded'? If the sound in both channels is identical, like mono sound (or the source located exactly midway between the microphones), do the hillsides then move out to each side, and come in from both sides, in identical moves? If so, the bottom of the V, the valley between the hillsides, forms a straight line, but moving up and down, and so does the stylus tip. Or, do the two hillsides move in parallel to the same side and back again, 'holding hands'? Then the stylus tip moves from side to side, but not up and down? On a 'true' monophonic record, the V track goes from side to side, with constant depth, suggesting that for a stereo record, the second alternative is the right one, but I am far from 100% sure about that. Can anyone tell, with absolute certainty, which alternative is right for a stereo record? Does the stylus tip go in a straight line, but up and down, or from side to side, but at the same vertical depth, when a stereo record plays exactly the same sound in both speakers?
I recall this: Digital Needle - A Virtual Gramophone[^]
-
Vinyl records, as well as 78s (with the exception of a very few, and incompatible, records from the very first years), have a V-formed track where the stylus rides. On a stereo record, the sound from the left channel is engraved in the 'hillside' that is the one arm of the V, the right channel sound is engraved in the opposite V arm hillside. The question: How is the sound 'encoded'? If the sound in both channels is identical, like mono sound (or the source located exactly midway between the microphones), do the hillsides then move out to each side, and come in from both sides, in identical moves? If so, the bottom of the V, the valley between the hillsides, forms a straight line, but moving up and down, and so does the stylus tip. Or, do the two hillsides move in parallel to the same side and back again, 'holding hands'? Then the stylus tip moves from side to side, but not up and down? On a 'true' monophonic record, the V track goes from side to side, with constant depth, suggesting that for a stereo record, the second alternative is the right one, but I am far from 100% sure about that. Can anyone tell, with absolute certainty, which alternative is right for a stereo record? Does the stylus tip go in a straight line, but up and down, or from side to side, but at the same vertical depth, when a stereo record plays exactly the same sound in both speakers?
Mono recording uses 90/180 degree wiggles. In other words, left to right wiggles. Stereo uses 45/225 degree wiggles for one channel and 315/135 degree wiggles for the other channel. So if a mono record is played on a stereo cartridge, the same signal appears in both channels. If a stereo recording has a perfectly centered sound, it would be (almost) 90/180 degree wiggles. This is one of the reasons why exact cartridge alignment makes such a difference in sound quality.
-
I recall this: Digital Needle - A Virtual Gramophone[^]
When I was a boy, in the 1970, our neighbor shipped some old, worn records across the pond to have them photographically read/restored - he got the results as an open-reel magnetic tape. Today, I am quite sure that must have been 78rpm records - I very much doubt that the technology of the 1970s was good enough to do it on 'microgroove' vinyl records. He was very satisfied with the results, claiming to hear instruments (he specifically mentioned triangles) that he had never heard from the mechanical record. The stylus rides down in track, not touching the edges between the 'land' and 'valley' surface areas, so these edges are not worn with use. In stray light almost parallel to the surface let the original, the original, unworn edge could be read with an optical pickup, as the borderline between the lit land and the dark, shadowed valley. I was considering if a similar technique would be possible on 'modern', stereo vinyl records. Some simple back-of-envelope calculations did tell me that it is not a project for your hobby workshop in your basement. Say that you use a microscope with an image area 0.2 to 0.1 mm wide - vinyl grooves are typically packed 7.5/mm, on some records up to 10/mm. Assume the microscope ocular is a high-res 8K video sensor. Since the image must capture both sides of the groove (on a stereo record), there are 4K pixels to each side. This is to capture the amplitude of the signal: 12 bits of resolution at the very most. More like 11 bits on the average. Maybe you can steal an extra bit of resolution by looking at the bit on the edge: is it almost white, light gray, middle gray, dark gray, almost black? What about the microscope optics? Can it, for an image area, 0.1 mm wide, provide a resolution of 8K pixels? That is pixel width of approx. 0.01 - 0.02 µm on the record surface. The traditional resolution measures, such as line pairs/mm is not directly transferable to sensor pixels, and I am by no means an expert in the field anyway, but from what I read about microscopes, that is significantly beyond the capabilities of traditional optical ones. The wavelength of the light limits the resolution. Maybe you can get closer by using short-wave UV? ... What fascinates me is that what I want to read out are no smaller details than the pickup stylus manages to read out! From a top quality, factory new vinyl record, it can deliver at least 60 dB S/N. That sure isn't more than 10 bits, but it does it almost without any effort. Why should it be so much more difficult to trace the edge
-
Very cool - never gave it much thought before. Just yesterday I brought a couple Weather Report albums circa 1970's up from the basement to play this week. :)
I had the Heavy Weather album, hat with lightning on the sleeve. Favorite tune, "Birdland". Just looked it up, it's the Indiana Jones Fedora!
-
When I was a boy, in the 1970, our neighbor shipped some old, worn records across the pond to have them photographically read/restored - he got the results as an open-reel magnetic tape. Today, I am quite sure that must have been 78rpm records - I very much doubt that the technology of the 1970s was good enough to do it on 'microgroove' vinyl records. He was very satisfied with the results, claiming to hear instruments (he specifically mentioned triangles) that he had never heard from the mechanical record. The stylus rides down in track, not touching the edges between the 'land' and 'valley' surface areas, so these edges are not worn with use. In stray light almost parallel to the surface let the original, the original, unworn edge could be read with an optical pickup, as the borderline between the lit land and the dark, shadowed valley. I was considering if a similar technique would be possible on 'modern', stereo vinyl records. Some simple back-of-envelope calculations did tell me that it is not a project for your hobby workshop in your basement. Say that you use a microscope with an image area 0.2 to 0.1 mm wide - vinyl grooves are typically packed 7.5/mm, on some records up to 10/mm. Assume the microscope ocular is a high-res 8K video sensor. Since the image must capture both sides of the groove (on a stereo record), there are 4K pixels to each side. This is to capture the amplitude of the signal: 12 bits of resolution at the very most. More like 11 bits on the average. Maybe you can steal an extra bit of resolution by looking at the bit on the edge: is it almost white, light gray, middle gray, dark gray, almost black? What about the microscope optics? Can it, for an image area, 0.1 mm wide, provide a resolution of 8K pixels? That is pixel width of approx. 0.01 - 0.02 µm on the record surface. The traditional resolution measures, such as line pairs/mm is not directly transferable to sensor pixels, and I am by no means an expert in the field anyway, but from what I read about microscopes, that is significantly beyond the capabilities of traditional optical ones. The wavelength of the light limits the resolution. Maybe you can get closer by using short-wave UV? ... What fascinates me is that what I want to read out are no smaller details than the pickup stylus manages to read out! From a top quality, factory new vinyl record, it can deliver at least 60 dB S/N. That sure isn't more than 10 bits, but it does it almost without any effort. Why should it be so much more difficult to trace the edge
-
When I was a boy, in the 1970, our neighbor shipped some old, worn records across the pond to have them photographically read/restored - he got the results as an open-reel magnetic tape. Today, I am quite sure that must have been 78rpm records - I very much doubt that the technology of the 1970s was good enough to do it on 'microgroove' vinyl records. He was very satisfied with the results, claiming to hear instruments (he specifically mentioned triangles) that he had never heard from the mechanical record. The stylus rides down in track, not touching the edges between the 'land' and 'valley' surface areas, so these edges are not worn with use. In stray light almost parallel to the surface let the original, the original, unworn edge could be read with an optical pickup, as the borderline between the lit land and the dark, shadowed valley. I was considering if a similar technique would be possible on 'modern', stereo vinyl records. Some simple back-of-envelope calculations did tell me that it is not a project for your hobby workshop in your basement. Say that you use a microscope with an image area 0.2 to 0.1 mm wide - vinyl grooves are typically packed 7.5/mm, on some records up to 10/mm. Assume the microscope ocular is a high-res 8K video sensor. Since the image must capture both sides of the groove (on a stereo record), there are 4K pixels to each side. This is to capture the amplitude of the signal: 12 bits of resolution at the very most. More like 11 bits on the average. Maybe you can steal an extra bit of resolution by looking at the bit on the edge: is it almost white, light gray, middle gray, dark gray, almost black? What about the microscope optics? Can it, for an image area, 0.1 mm wide, provide a resolution of 8K pixels? That is pixel width of approx. 0.01 - 0.02 µm on the record surface. The traditional resolution measures, such as line pairs/mm is not directly transferable to sensor pixels, and I am by no means an expert in the field anyway, but from what I read about microscopes, that is significantly beyond the capabilities of traditional optical ones. The wavelength of the light limits the resolution. Maybe you can get closer by using short-wave UV? ... What fascinates me is that what I want to read out are no smaller details than the pickup stylus manages to read out! From a top quality, factory new vinyl record, it can deliver at least 60 dB S/N. That sure isn't more than 10 bits, but it does it almost without any effort. Why should it be so much more difficult to trace the edge
Hmmm, You are describing something very similar to [Project IRENE](https://irene.lbl.gov/) from Berkeley Lab. Apparently there is a working model here in the U.S. at the Library of Congress. [Project IRENE: Analyzing Images to Digitize Sound on Historic Audio Recordings | Library of Congress](https://www.loc.gov/item/webcast-7555) Looks interesting.
-
Hmmm, You are describing something very similar to [Project IRENE](https://irene.lbl.gov/) from Berkeley Lab. Apparently there is a working model here in the U.S. at the Library of Congress. [Project IRENE: Analyzing Images to Digitize Sound on Historic Audio Recordings | Library of Congress](https://www.loc.gov/item/webcast-7555) Looks interesting.
I am not yet through the entire 1 hr video, but it is running right now. Vax cylinders (as well as couple disc variants from around 1900-1910) create a much larger problem than vinyl records, waving from side to side: The groove is straight, and the depth is varying. I heard the lady in the video mention that the variation in depth is, for the vax cylinders, about 10 microns. I am really curious to learn how you can optically read depth variations over a span of 10 µm in a straight track with any precision! That appears to be a much more difficult task than reading waving land/valley edge of a vinyl record!
-
Fascinating solution! However, the guy in the video seems very focused on 'no digitalization', direct to analog. That means you have no option to use digital techniques to e.g. handle scratches, dust particles etc. The video doesn't tell anything about how the laser beam is read. The simple 2D figure shows two laser beams perpendicular to the sides, but how do you then detect how far away the side is? How is the amplitude measured? I wouldn't be surprised if that is a business secret ...
-
I had the Heavy Weather album, hat with lightning on the sleeve. Favorite tune, "Birdland". Just looked it up, it's the Indiana Jones Fedora!
Yep, first track on side 1. My favorite tune of theirs as well. I turn up the volume for that one. Their albums have great cover art.
-
Fascinating solution! However, the guy in the video seems very focused on 'no digitalization', direct to analog. That means you have no option to use digital techniques to e.g. handle scratches, dust particles etc. The video doesn't tell anything about how the laser beam is read. The simple 2D figure shows two laser beams perpendicular to the sides, but how do you then detect how far away the side is? How is the amplitude measured? I wouldn't be surprised if that is a business secret ...